基于深度学习与不平衡样本集的输电线路故障分类

针对输电线路各类型故障样本间的数量不平衡会造成人工智能算法对故障中的少数类样本识别精度不足的问题,提出了一种基于Borderline-SMOTE（BSMOTE）算法与卷积神经网络（CNN）相结合的输电线路故障分类方法。该方法首先利用BSMOTE算法对位于分类边界上的少数类样本进行过采样合成处理,改善样本间的不平衡度,然后将所提取的一维故障电流信号样本重构成二维灰度图像数据形式,并在Pytorch深度学习框架下搭建了CNN网络模型,利用模型的自主学习能力对灰度图像进行特征自提取与辨识,减少传统人工设计特征提取的工序,完成对输电线路故障类型的分类。实验结果表明该模型能够提高对少数类故障样本的识别能力,准确地判断故障类型,并对噪音具有较强的抗干扰能力。     

考虑客户画像和能量枢纽的配电网多目标抢修重构模型

随着大规模风电以及能量枢纽接入配电网,配电网故障后抢修重构面临新挑战,提出一种考虑客户画像和能量枢纽的配电网多目标多故障抢修重构模型。通过对配电网用电客户多因素精准画像,确定权重,建立兼顾客户画像和能量枢纽的配电网多目标抢修重构模型,提出一种改进NSGA-II算法以适应对模型求解,从而获得Pareto解集,并利用模糊最大满意度进行决策。算例仿真表明所提模型正确有效,与传统模型相比更加精细而客观。     

基于CPSO-BP神经网络的风电并网暂态电压稳定评估

针对目前传统方法难以快速、准确判断风电并网后系统暂态电压稳定性的问题,提出了一种基于CPSO-BP组合的风电并网暂态电压稳定评估方法。首先采用混沌理论对粒子群算法的不足进行改善,应用改进后的算法对神经网络的初始权值和阈值进行优化,然后利用系统故障前后采集的传统物理量和风电场相关的物理量作为BP神经网络输入特征量进行监督学习,最后将训练得到的模型应用于风电并网系统的暂态电压稳定评估中。利用英格兰10机39节点系统标准算例进行风电并网仿真分析,结果证明了所提方法的有效性。     

基于注意力机制和卷积神经网络的异步电动机三相电压不平衡损耗研究

配电网中三相电压不平衡对异步电动机损耗会造成较大影响。运用等效电路公式分析三相电压不平衡影响下电动机损耗存在精度不稳定、需要参数过多且数学模型过于复杂等问题。针对以上问题,提出一种基于注意力机制和卷积神经网络(CNN)的异步电动机损耗评估方法。该方法将实测电机数据作为输入,引入注意力机制为输入特征赋予不同权重;采用卷积层和全连接层组成的CNN构架对异步电动机实测数据进行学习,最后完成损耗评估。以现场试验得到的电机损耗数据作为实际算例,该方法评估损耗与实测损耗平均误差仅为0.717%和0.549%,并与其他典型机器学习算法进行对比,结果表明所提方法具有更好的损耗评估性能。     

基于卷积神经网络的直线同步电动机电枢绕组故障诊断

通过绕组函数理论对直线同步电动机进行分析,提出一种基于卷积神经网络(CNN)的直线同步电动机故障诊断方法。从直线同步电动机的数学模型出发,基于绕组函数理论对电动机正常状态和匝间短路故障状态进行仿真,对电流波形图进行快速傅里叶变换(FFT)得到不同状态的数据集。利用CNN中的GoogLeNet网络结构,在保持网络空间维度的同时不增加故障诊断的计算量。将数据集输入到网络模型进行故障诊断,仿真结果表明GoogLeNet网络结构对直线同步电动机电枢绕组的短路故障识别率达到了96.5%以上。     

基于电流变化的环状直流配电网故障定位方法

针对现有故障测距方法无法同时对单极接地故障和极间故障进行定位,以及利用电容放电阶段进行故障定位时忽略了多端换流站都会对故障点注入电流等问题,提出一种利用故障线路上的全电流分量对多端柔性直流配电网进行故障定位的方法。通过联立与故障点相邻两网孔的电压方程,从算法上解决了过渡电阻对求解故障位置的影响。在simulink中搭建6端环状模型对所提方法进行仿真,结果验证所提方法可以在发生线路故障时能快速识别故障并进行可靠定位,满足直流配电网的故障测距要求。     

基于高频暂态稀疏测量的复杂直流配电网故障定位

为了解决直流配电网中高速通信测量点不足、故障特征复杂等导致的故障定位精度差的问题,提出了一种适用于复杂直流配电网的新型直流极间短路故障定位算法。首先,根据高频瞬态电流环路,构建含有电平转换器和DC/DC转换器的高频阻抗等效模型为故障过程提供稳定的阻抗值。其次,利用BCS理论推导与稀疏测量点相对应的节点高频暂态电压方程。最后,结合节点高频瞬态电压方程和贝叶斯压缩感知理论,求解节点高频瞬态电流稀疏矢量实现故障定位。实验结果表明:所提算法对测量点的数量要求较低,不需要严格同步地测量数据,且不受转换器的控制策略和过渡电阻影响,具有较高的故障定位准确性。     

Gap Junctional Communication via Connexin43 between Purkinje Fibers and Working Myocytes Explains the Epicardial Activation Pattern in the Postnatal Mouse Left Ventricle

The mammalian ventricular myocardium forms a functional syncytium due to flow of electrical current mediated in part by gap junctions localized within intercalated disks. The connexin (Cx) subunit of gap junctions have direct and indirect roles in conduction of electrical impulse from the cardiac pacemaker via the cardiac conduction system (CCS) to working myocytes. Cx43 is the dominant isoform in these channels. We have studied the distribution of Cx43 junctions between the CCS and working myocytes in a transgenic mouse model, which had the His-Purkinje portion of the CCS labeled with green fluorescence protein. The highest number of such connections was found in a region about one-third of ventricular length above the apex, and it correlated with the peak proportion of Purkinje fibers (PFs) to the ventricular myocardium. At this location, on the septal surface of the left ventricle, the insulated left bundle branch split into the uninsulated network of PFs that continued to the free wall anteriorly and posteriorly. The second peak of PF abundance was present in the ventricular apex. Epicardial activation maps correspondingly placed the site of the first activation in the apical region, while some hearts presented more highly located breakthrough sites. Taken together, these results increase our understanding of the physiological pattern of ventricular activation and its morphological underpinning through detailed CCS anatomy and distribution of its gap junctional coupling to the working myocardium.     

Auxin and Root Gravitropism: Addressing Basic Cellular Processes by Exploiting a Defined Growth Response

Root architecture and growth are decisive for crop performance and yield, and thus a highly topical research field in plant sciences. The root system of the model plant Arabidopsis thaliana is the ideal system to obtain insights into fundamental key parameters and molecular players involved in underlying regulatory circuits of root growth, particularly in responses to environmental stimuli. Root gravitropism, directional growth along the gravity, in particular represents a highly sensitive readout, suitable to study adjustments in polar auxin transport and to identify molecular determinants involved. This review strives to summarize and give an overview into the function of PIN-FORMED auxin transport proteins, emphasizing on their sorting and polarity control. As there already is an abundance of information, the focus lies in integrating this wealth of information on mechanisms and pathways. This overview of a highly dynamic and complex field highlights recent developments in understanding the role of auxin in higher plants. Specifically, it exemplifies, how analysis of a single, defined growth response contributes to our understanding of basic cellular processes in general.     

Method for discrimination of false targets in multistation radar systems based on the deep neural network

For the existing jamming discrimination methods for multistation radar systems,only the single feature of target echo space correlation is utilized as the metric,which leads to insufficient comprehensiveness of feature extraction,so that effectiveness and universality are insufficient for the discrimination algorithm.In this paper,an identification method in multistatic radar systems based on the deep neural network is proposed.This method combines the characteristics of multistatic radar systems cooperative detection technology,which has many available resources and strong scheduling ability in space,time and frequency domain,with the strong model learning and feature representation ability in the process of information processing on the deep neural network,so that it can effectively apply to the field of anti-deception jamming.Full use is made of unknown information about echo data to obtain more multi-dimensional,more comprehensive,more complete and deeper feature differences besides correlation,so as to achieve a better jamming discrimination effect.Simulation results show that the proposed method can effectively reduce the influence of noise and pulse number on the jamming discrimination performance.At the same time,the limitation of the target echo correlation coefficient on anti-jamming technology under nonideal conditions is alleviated,which broadens the boundary conditions of the application process.